Reference to Video AI uses approved images to guide a recognizable character, object, product, or visual identity into a new moving scene. Add clear references, describe the setting and action, then compare models that support reference-led generation.
Reference images give the model visual context for the character, object or product in your prompt.
Try it nowDescribe a fresh setting and action while the references guide what should remain recognizable.
Try it nowThe model picker is filtered to modes that accept reference-led video generation.
Try it nowReference guidance is designed for recognizable subjects in newly described scenes. Use Image to Video AI when one prepared frame should anchor the composition, or Motion Control when a recorded performance should supply the movement.
Try it nowUse clear reference images to guide a character, object, mascot, or product appearance while generating a scene that did not exist in the source material.
Develop lifestyle and campaign concepts around product imagery you own, while the prompt defines the setting, action, atmosphere, and camera treatment.
Where the chosen model provides multiple reference slots, add compatible views or subjects and let the visible composer define the current limit.
Test the same approved assets and scene brief with compatible models to compare subject continuity, movement, composition, and prompt interpretation.
Carry defining visual traits into separate scene concepts for story planning, campaign exploration, mascot content, or character design reviews.
Use Image to Video AI when one exact frame should lead the motion, or Video to Video AI when existing footage should guide timing and movement.
Use approved product, mascot, or campaign references to explore connected scenes without describing every visual trait again in each brief.
Carry a recurring character into new environments and actions while preserving recognizable features that help separate shots feel related.
Build lifestyle motion concepts around product photographs you are authorized to use, then review whether important product details remain accurate.
Test how an original character reads in different scenes, poses, and moods before investing in a longer animation or production sequence.
Upload clear views of the subject you want the model to follow.
Write the action, setting and camera direction for the video.
Select a compatible model, review its inputs and create the result.
Use clean references with the subject clearly visible against a simple background.
Refer to the subject the same way in your prompt every time you generate.
Only use references you own or are allowed to use.
Choose complementary views with consistent styling and avoid references that contradict each other on defining colors, shapes, or clothing.
Review faces, logos, text, and product geometry in the output because reference guidance improves continuity but does not guarantee exact reproduction.
Reference to Video AI generates a new scene while using supplied images as visual guidance for a subject. It is useful when a character, product, object, or mascot should remain recognizable even though the action and environment come from a new prompt.
Use clear images of a character, object, or product that you own or are allowed to use. The number and type of reference slots depend on the selected model, and the composer shows the exact inputs available for the current mode.
No. Reference to Video AI provides visual guidance, but pose changes, occlusion, fast action, and complex scenes can alter fine details. Review faces, text, logos, clothing, and product geometry in every result, especially when accurate brand or character representation matters.
Generation uses pay-as-you-go credits rather than a fixed price for every task. The estimate beside the Generate button updates for the model and parameters you select, so you can review the expected charge before submitting. Changing the model, quality, format, or other available controls can change that estimate.
You can explore the page, compare the available models, and inspect their controls without signing in. An account is required when you are ready to submit a generation task, monitor its progress, and access the completed result in your workspace.
Requirements depend on the selected model and workflow. The composer displays the inputs, accepted upload types, and controls that apply to your current selection. If you switch models, review those fields again because supported assets and settings may differ.
Commercial use depends on your rights to every input, the selected model terms, and the rules that apply where you publish. Use only prompts and source assets you are authorized to use, review the applicable terms, and check the final clip before distributing it.
Add reference images you are authorized to use, describe the new action and setting, review the model estimate, and generate above.
Try it now
